Catch up on select AI news and developments from the past week or so:
OpenAI is expected to release GPT-5 this summer (or thereabouts). The big takeaway: the new version will be more autonomous, relying less on human input, expanding its scope beyond text generation to more complex tasks. Think advancements in data analysis, report writing, email management, coding, and more. OpenAI is now concluding the training of GPT-5 and preparing safety testing to ensure reliability before its release.
Microsoft has hired Mustafa Suleyman, a co-founder of AI startup Inflection and Google's DeepMind, to lead its consumer AI initiatives. He will oversee Copilot, Bing, and Edge, among other projects. Also joining Microsoft is another Inflection co-founder, Karén Simonyan. Microsoft is competing with Google in AI. Microsoft has invested $10 billion in ChatGPT-creator OpenAI. It has also been working to integrate AI into its own products.
Apple and Google may partner on AI. Apple is apparently in early talks with Google to license the latter's Gemini AI engine for use in the iPhone. That would allow Apple to catch up with competitors, including Microsoft, and develop AI-powered features for its products. Google, meanwhile, would gain access to Apple's user base, thus promote its AI technology.
YouTube has issued new rules for AI-generated content. It's requiring creators to disclose the use of AI-generated content in videos. The disclosure is intended to provide viewers with transparency about the authenticity of the content they are watching, particularly content that appears realistic but has been digitally altered or created using AI. The new policy is intended to help curb AI-generated misinformation, particularly in a US presidential election year. The policy does not apply to AI-generated animations aimed at children or to "minor" edits that are primarily aesthetic.
The Securities and Exchange Commission has started clamping down on investment advisers who falsely claim to use AI in providing their services. "AI washing" refers to companies' untrue claims of AI use—echoing the term "green-washing," or misleading consumers about environment-conscious policies.
The BBC is considering building an in-house AI model. The national broadcaster is drawing up plans to build and train its own AI model using its text archives. The output would be for internal use only. The BBC is holding talks with AI tech firms to sell access to its archives to train their AI models.
Musk's xAI has open-sourced Grok. Developers and researchers can now use the "weights and architecture" that power X’s Grok chatbot for free in their own projects. An open release is available on GitHub. xAI's Grok plans to compete with OpenAI, Meta, Google, and others.
The US is pushing global AI regulations. A US-led United Nations General Assembly resolution seeks to establish global consensus on AI. The intent is, ostensibly, "to promote safe and trustworthy AI systems in a push to increase global regulation and sustainable development of the technology."
Useful Tidbit: From AI to Z: A Starter Guide to Using Generative AI in PR & Comms is available for download from Cision (reg. required).
You can find last week's AI Update here.
